基于多智能体 LLM 与 RAG 优化甲状腺功能亢进入院记录中现病史质量的研究
Quality optimization with low positive factual hallucination for the HPI in hyperthyroidism admission notes using multi agent LLM with RAG
一项研究开发了针对甲状腺功能亢进的入院记录质量控制模型 HQ-ANR,该模型结合多智能体协作和诊断驱动检索增强生成技术优化现病史,基于 Qwen3-32B 构建并锚定双源知识库。在盲评 30 份住院医书写的甲亢记录中,该模型将中位 PDQI-9 评分从 39.00 提高至 42.00,将正向事实幻觉率从 26.67% 降至 6.67%,平均减少 112.9 秒的修订时间。
- Article
- Open access
- Published: 07 October 2026
- Qingya Zeng1,2 na1,
- Huiyu Lan1,2 na1,
- Hang Lv3 na1,
- Jing Lin ORCID: orcid.org/0009-0003-6470-13831,2,
- Pengxiang Zhan3,
- Binteng Cai3,
- Yacong Yang3,
- Ji Huang4,5,
- Lan Lin4,5,
- Yan Chen1,2,
- Xiongfeng Chen ORCID: orcid.org/0000-0002-7445-36731,2,
- Wei Lin ORCID: orcid.org/0000-0002-0455-87521,2,
- Yanchao Tan3 &
- …
- Gang Chen1,2
npj Digital Medicine (2026) Cite this article
We’re sharing this article early to provide faster access to peer-reviewed, accepted research. It is citable and carries a permanent DOI. This version is subject to further edits and will be replaced automatically by the final Version of Record. All legal disclaimers apply.
Abstract
Admission note quality affects clinical safety and research reliability, yet resident drafts often suffer from missing information, unclear logic, and other issues, requiring time-consuming manual review. Current large language models (LLMs) focus on text generation or formatting correction but rarely optimize clinical logic or mitigate hallucinations in specialized notes. We developed HQ-ANR (High-Quality Admission Note Refinement), a hyperthyroidism-specific quality-control model that integrates multi-agent collaboration with diagnosis-driven retrieval-augmented generation (RAG) to optimize the history of present illness (HPI) in admission notes. Built on a general-purpose LLM (Qwen3-32B), HQ-ANR anchors generation to a dual-source knowledge base (guidelines and expert notes) and employs specialized agents for symptoms, examinations, and treatments. In a blinded evaluation of 30 resident-drafted hyperthyroidism notes, HQ-ANR significantly improved median Physician Documentation Quality Instrument (PDQI-9) scores (from 39.00 to 42.00) and reduced the hallucination rate (positive factual hallucinations only) to 6.67%, compared to 26.67% for a general LLM with chain-of-thought (CoT) prompting. Efficiency analysis showed HQ-ANR reduced physician refinement time by an average of 112.9 s per note (p < 0.001). These results demonstrate that augmenting a general LLM with external knowledge and a multi-agent framework can achieve reliable, in-depth optimization of clinical notes without expensive vertical training, offering a practical pathway toward intelligent, safe documentation quality control and reducing physician documentation burden.
Similar content being viewed by others



Acknowledgements
The authors thank the clinical and informatics staff of Fuzhou University Affiliated Provincial Hospital for assistance with data extraction and de-identification, and acknowledge the financial support provided by Noncommunicable Chronic Diseases-National Science and Technology Major Project (Grant No. 2024ZD0531600), Department of Endocrinology (National Clinical Key Specialty, Grant No. 2023001), Fujian Research and Training Grants for Young and Middle-aged Leaders in Healthcare (Grant No. (2023)417#), the Natural Science Foundation of Fujian Province (Grant No. 2022J011017), Fujian Joint Fund for Scientific and Technological Innovation Projects (Grant No. 2023Y9295), and Fujian Clinical Center of Endocrinology and Metabolism (Grant No. 006112604). The funders played no role in study design, data collection, analysis and interpretation of data, or the writing of this manuscript.
Author information
Author notes
These authors contributed equally: Qingya Zeng, Huiyu Lan, Hang Lv.
Authors and Affiliations
Shengli Clinical Medical College, Fujian Medical University, Fuzhou, PR China
Qingya Zeng, Huiyu Lan, Jing Lin, Yan Chen, Xiongfeng Chen, Wei Lin & Gang Chen
Department of Endocrinology, Fuzhou University Affiliated Provincial Hospital, Fuzhou, PR China
Qingya Zeng, Huiyu Lan, Jing Lin, Yan Chen, Xiongfeng Chen, Wei Lin & Gang Chen
College of Computer and Data Science, Fuzhou University, Fuzhou, PR China
Hang Lv, Pengxiang Zhan, Binteng Cai, Yacong Yang & Yanchao Tan
Information Management Center, Fuzhou University Affiliated Provincial Hospital, Fuzhou, PR China
Ji Huang & Lan Lin
Fujian Provincial Key Laboratory of Medical Big Data Engineering, Fuzhou, PR China
Ji Huang & Lan Lin
Authors
- Qingya Zeng
- Huiyu Lan
- Hang Lv
- Jing Lin
- Pengxiang Zhan
- Binteng Cai
- Yacong Yang
- Ji Huang
- Lan Lin
- Yan Chen
- Xiongfeng Chen
- Wei Lin
- Yanchao Tan
- Gang Chen
Corresponding authors
Correspondence to Wei Lin, Yanchao Tan or Gang Chen.
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Zeng, Q., Lan, H., Lv, H. et al. Quality optimization with low positive factual hallucination for the HPI in hyperthyroidism admission notes using multi agent LLM with RAG. npj Digit. Med. (2026). https://doi.org/10.1038/s41746-026-03321-x
Received: 25 December 2025
Accepted: 16 September 2026
Published: 07 October 2026
DOI: https://doi.org/10.1038/s41746-026-03321-x
来源:Nature 健康护理主题 · nature.com